So, this is the current result with my Dropsnow-Macros.

First, I sampled a few thousand points onto the object (roughly 11.000 or
so).
Then, I discarded particles which had too much roof above them, cause
snowflakes wouldn't be possible to reach those spots.
Then, I checked the close vicinity of every particle with trace()-calls to
the object. I check how large the radius for the particle can be, how high
it can possibly get, if there's a wall close-by etc. These measurements
affect how the particle will be scaled afterwards to simulate stacking along
walls.
At the end, every particle is visualised with a cylindrical blob-element.

Works pretty well so far, though it needs some more tweakage of course. This
animation doesn't look photorealistic or such, but should give a sense of
what might be possible with the macros.

Also note the single particle at the corner where the cylinder is paced.
With the cylinder close by it is treated like a high-stacked particle, but
as it is standing alone, the blob doesn't add up enough to actually touch
the cylinder, and thus, I get a pillar... I guess that for cases like these,
a manual refinement (e.g. by clipping such a particle away) is best, rather
than try to invent some heavy-duty parsing script to get rid of single
particles out of thousand-bunches...

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>

Post a reply to this message

Attachments:
Download 'test_snow_2.mpg' (283 KB)